fix: update noodles-group to 0.97 (#678)

Author: tedil

Cargo.lock

@@ -59,7 +59,7 @@ tempfile = "3.19.1"
 lru = "0.13.0"
 
 [dependencies.noodles]
-version = "0.77.0"
+version = "0.97.0"
 features = ["bed", "bgzf", "core", "csi", "gff", "tabix", "vcf"]
 
 [build-dependencies]

Cargo.toml

@@ -5,6 +5,7 @@ use std::{fs::File, io::BufReader, path::PathBuf};
 use clap::Parser;
 
 use indicatif::ParallelProgressIterator;
+use noodles::core::Position;
 use rayon::prelude::*;
 
 use crate::common::{self, cli::extract_chrom, keys, spdi};
@@ -59,12 +60,14 @@ fn copy_cf_bed(
 ) -> Result<(), anyhow::Error> {
     let mut reader = File::open(path_bed)
         .map(BufReader::new)
-        .map(noodles::bed::Reader::new)?;
+        .map(noodles::bed::Reader::<3, _>::new)?;
 
     tracing::info!("  reading BED records...");
-    let bed_records = reader
-        .records::<3>()
-        .collect::<Result<Vec<noodles::bed::Record<3>>, _>>()?;
+    let mut bed_records = Vec::new();
+    let mut record = noodles::bed::Record::default();
+    while reader.read_record(&mut record)? != 0 {
+        bed_records.push(record.clone());
+    }
     tracing::info!(
         "  will process {} BED records in parallel...",
         bed_records.len()
@@ -75,9 +78,15 @@ fn copy_cf_bed(
         .progress_with(common::cli::progress_bar(bed_records.len()))
         .map(|record| {
             let chrom = record.reference_sequence_name();
-            let start: usize = record.start_position().into();
+            let start: usize = record
+                .feature_start()
+                .unwrap_or(Position::new(1).unwrap())
+                .into();
             let start = start + 1;
-            let stop: usize = record.end_position().into();
+            let stop: usize = record
+                .feature_end()
+                .unwrap_or(Ok(Position::new(1).unwrap()))?
+                .into();
 
             let start = spdi::Pos {
                 sequence: chrom.to_string(),

src/db_utils/cli/copy.rs

@@ -59,6 +59,7 @@ fn tsv_import(
         .reference_sequence_names()
         .iter()
         .filter_map(|chrom| {
+            let chrom = &chrom.to_string();
             let canon_chrom = chrom.strip_prefix("chr").unwrap_or(chrom);
             if common::cli::is_canonical(canon_chrom) {
                 Some((common::cli::canonicalize(canon_chrom), chrom.clone()))

src/dbsnp/cli/import.rs

@@ -122,6 +122,7 @@ pub fn build_windows(
             .reference_sequence_names()
             .iter()
             .filter_map(|chrom| {
+                let chrom = &chrom.to_string();
                 let canon_chrom = chrom.strip_prefix("chr").unwrap_or(chrom);
                 if common::cli::is_canonical(canon_chrom) {
                     Some((common::cli::canonicalize(canon_chrom), chrom.clone()))

src/freqs/cli/import/mod.rs

@@ -106,11 +106,11 @@ fn gff_import(
     // Import of RefSeq GFF data.
     let contig_map = ContigMap::new(args.genome_release.into());
     let mut reader = gff::Reader::new(std::io::BufReader::new(reader));
-    for result in reader.records() {
+    for result in reader.record_bufs() {
         let record = result?;
 
         // Resolve reference sequence name to contig name (for canonical ones).
-        let seq_name = record.reference_sequence_name();
+        let seq_name = &record.reference_sequence_name().to_string();
         let chromosome = match contig_map.chrom_name_to_seq(seq_name) {
             Ok(sequence) => {
                 if is_canonical(&sequence.name) {
@@ -129,14 +129,15 @@ fn gff_import(
         };
 
         /// Helper function to extract a key from the attributes of a record.
-        fn extract(record: &gff::Record, key: &str) -> Result<String, anyhow::Error> {
+        fn extract(record: &gff::feature::RecordBuf, key: &str) -> Result<String, anyhow::Error> {
+            let key_ = key.as_bytes();
             let value = record
                 .attributes()
-                .get(key)
+                .get(key_)
                 .ok_or_else(|| anyhow::anyhow!("problem with {} attribute: {:?}", key, record))?;
             match value {
-                gff::record::attributes::field::Value::String(s) => Ok(s.clone()),
-                gff::record::attributes::field::Value::Array(arr) => {
+                gff::feature::record_buf::attributes::field::Value::String(s) => Ok(s.to_string()),
+                gff::feature::record_buf::attributes::field::Value::Array(arr) => {
                     if arr.is_empty() {
                         Err(anyhow::anyhow!(
                             "problem with {} attribute: {:?}",
@@ -147,7 +148,7 @@ fn gff_import(
                         if arr.len() > 1 {
                             tracing::warn!("multiple values for {} attribute: {:?}", key, record);
                         }
-                        Ok(arr[0].clone())
+                        Ok(arr[0].clone().to_string())
                     }
                 }
             }

src/functional/cli/import.rs

@@ -65,7 +65,8 @@ fn tsv_import(
         .reference_sequence_names()
         .iter()
         .filter_map(|chrom| {
-            let canon_chrom = chrom.strip_prefix("chr").unwrap_or(chrom);
+            let chrom = chrom.to_string();
+            let canon_chrom = chrom.strip_prefix("chr").unwrap_or(&chrom);
             if common::cli::is_canonical(canon_chrom) {
                 Some((common::cli::canonicalize(canon_chrom), chrom.clone()))
             } else {

src/gnomad_mtdna/cli/import.rs

@@ -124,7 +124,8 @@ fn vcf_import(
         .reference_sequence_names()
         .iter()
         .filter_map(|chrom| {
-            let canon_chrom = chrom.strip_prefix("chr").unwrap_or(chrom);
+            let chrom = chrom.to_string();
+            let canon_chrom = chrom.strip_prefix("chr").unwrap_or(&chrom);
             if common::cli::is_canonical(canon_chrom) {
                 Some((common::cli::canonicalize(canon_chrom), chrom.clone()))
             } else {

src/gnomad_nuclear/cli/import.rs

@@ -54,7 +54,8 @@ fn tsv_import(
         .reference_sequence_names()
         .iter()
         .filter_map(|chrom| {
-            let canon_chrom = chrom.strip_prefix("chr").unwrap_or(chrom);
+            let chrom = chrom.to_string();
+            let canon_chrom = chrom.strip_prefix("chr").unwrap_or(&chrom);
             if common::cli::is_canonical(canon_chrom) {
                 Some((common::cli::canonicalize(canon_chrom), chrom.clone()))
             } else {

src/helixmtdb/cli/import.rs

@@ -17,10 +17,7 @@ fn resolve_region(
 ) -> std::io::Result<usize> {
     header
         .reference_sequence_names()
-        .get_index_of(
-            std::str::from_utf8(region.name())
-                .map_err(|e| std::io::Error::new(std::io::ErrorKind::InvalidData, e))?,
-        )
+        .get_index_of(region.name())
         .ok_or_else(|| {
             std::io::Error::new(
                 std::io::ErrorKind::InvalidInput,
@@ -137,7 +134,7 @@ pub fn tsv_import(
             .filter(|(chrom, _, _)| {
                 header
                     .reference_sequence_names()
-                    .get_index_of(chrom)
+                    .get_index_of(chrom.as_bytes())
                     .is_some()
             })
             .map(|(chrom, begin, end)| {